
Early social media use linked to lower maths and reading scores, equivalent to six months of lost learning, Italian study finds
Italian pupils who opened social media accounts at age 11 or 12 scored significantly lower in maths and Italian by age 16 than peers who waited until 14 or 15, according to a study of over 5,000 students published in Nature Human Behaviour.
A new study from the University of Milano-Bicocca provides some of the strongest evidence yet that opening social media accounts at a young age is associated with worse academic performance. The research, published on 3 August in Nature Human Behaviour, tracked more than 5,200 Italian schoolchildren and found that those who created their first account at the start of secondary school fell behind in maths and Italian by the equivalent of about six months of schooling compared with those who delayed.
The study design
The research team, led by Professor Marco Gui, analysed surveys on social media habits and standardised test results in Italian, maths and English from 5,227 pupils across 28 schools in northern Italy. Data were collected between 2023 and 2024. Crucially, the researchers could observe students' academic performance before they began using social media, allowing them to verify that both groups were on similar trajectories beforehand.
It is important to underline that, being able to observe students' results before they started using social media, we were able to verify whether the groups were already following similar academic paths. This considerably strengthens the credibility of the causal interpretation.
The team also controlled for family structure, parents' education level, and socioeconomic background to isolate the effect of early social media adoption.
The performance gap
Pupils who opened their first social media account at age 11 or 12 scored lower in maths and Italian tests at age 13 to 14 than those who waited until at least age 14. By age 15 to 16, the gap in Italian remained similar while the maths gap widened further. The difference was quantified as roughly equivalent to six months of schooling. The benefit to those who delayed was about half the size of gains typically achieved through intensive, costly tutoring.
I think we now know there is a problem and that it's a complex problem, to do with how these platforms are built. Our research shows that at least during pre-adolescence, these social media are problematic.
English as a foreign language was the exception: no performance gap appeared between early and late adopters. The researchers hypothesised that the pervasive presence of English-language content on social media platforms may provide incidental exposure and informal practice.
Why the effect occurs
One of the strongest predictors of the performance gap was how often children checked their phones throughout the day and night. The researchers speculate that a continuous state of alert, driven by constant notifications and the impulse to check updates, acts as a persistent distraction that undermines concentration and learning.
Among all the possible explanations tested in our work, the most plausible is that what affects grades is the impact, over the course of the day, of continually checking social media, with consequent hyperstimulation and prolonged distraction.
Earlier research, cited by the authors, also points to reduced sleep quality, cognitive overload from the volume of information, and the simple displacement of study time as potential mechanisms, though these were not directly tested in the current study.
Policy context
The findings land amid a wave of legislative action across Europe. France is set to become the first EU country to ban social media access for under-15s after the summer holidays. Germany, Spain, Greece and Austria are also working on age-based restrictions for minors. The debates have so far focused mainly on mental health and addiction potential, but the new study adds academic performance to the list of concerns.
Klaus Zierer, professor of school pedagogy at the University of Augsburg, described the study as methodologically one of the most interesting published in this context, noting that the correlations it reveals are not random. The research group included collaborators from the University of Brescia, the Studi Socialis centre, and the Sloworking association.
- Mathematics
- 6 months of learning lost
- Italian
- 6 months of learning lost
- English
- 0 months of learning lost
